Salesforce Pushes 'Happy Middle' AI Approach Amid Enterprise Adoption Slowdown

Salesforce Chairman and CEO Marc Benioff spoke at The Six Five Summit: AI Unleashed 2026, discussing the current state of enterprise artificial intelligence. He stated that while consumer excitement around AI is high, large companies are only just beginning to implement it in their businesses.

Benioff emphasized that enterprise AI adoption has been slower than expected due to a 'tale of two cities' phenomenon: on one side are the enthusiastic media and public, while on the other are companies still relying on traditional software stacks. He met with over 200 customers during his recent eight-week trip to Europe and found many to be cautiously testing AI.

The company is pushing for a 'happy middle' approach, where deterministic enterprise systems are combined with probabilistic AI tools. This would allow businesses to unlock value from their existing investments while preserving security, governance, and control. Salesforce has already implemented this model in its own business and in customer environments, with examples including the 5 million customer service issues resolved by Casey and tens of millions of Slack users supported by Slackbot.

Benioff also discussed the company's hiring of thousands of new graduates to help build governed AI systems. He argued that while AI may change job requirements, it will not eliminate the need for software talent. Instead, enterprises will require people who can connect data, applications, and interfaces into secure systems.